Injection Molding - Median individual control charts

[ntmhx]

Hello everybody,

regarding the median individual control charts applied in a multicavity molding process each subgroup must contain measurements from all cavities? In order to construct a chart for 32 mold cavity we need 25 subgroups of 32 measuremens each?

thank you for your help

 

[Danny Hoover]

Re: Median individual control charts injection molding

Good morning,
We track ours by each cavity, not as a combination.
You still need samples from each cavity so the qty. is not less, just not combined
Hope this helps you
DH

 

[Statistical Steven]

Re: Median individual control charts injection molding

Good morning,
We track ours by each cavity, not as a combination.
You still need samples from each cavity so the qty. is not less, just not combined
Hope this helps you
DH

I have found that doing control charts for each cavity minimizes the need to get samples from all cavities. Usually we pick a representative sample of the cavities and control chart them.

 

[Bev D]

Re: Median individual control charts injection molding

a median chart is a great approach to things like multi-cavity molds or multi- head filling machines etc. (the differences between each cavity are 'fixed' and therefore if the subgroup includes parts across the cavities the within subgroup variation (Range or SD) will be incorrect and a violation of the requirement for a homogenous process. The median overcomes this.

depending on which median chart you use will most likely be constrained by the Median A2 value (commonly found tables only go up to 10) unless of course you are using software that does this.

you could chart the median using a moving range chart and avoid the A2 thing altogether...

or you could chart representative cavities (separately) as suggested.

 

[ntmhx]

...so you suggest to do individual charts for selected cavities not all of them? e.g form a 32 cavity mold to pick 5 and construct 5 individuals charts?

 

[Bev D]

...so you suggest to do individual charts for selected cavities not all of them? e.g form a 32 cavity mold to pick 5 and construct 5 individuals charts?

This is a good approach. I suggest taking the most vulnerable cavities to plot. You may be able to use a traditional Xbar r chart and subgroup by 3-5 sequential parts from each of the chosen cavities OR if the process makes it difficult to get sequential parts or the measurements are difficult an individuals chart will work.

You will have to play with the data...

 

[ntmhx]

Thank you, I'll try your suggestion. Since recently I was collecting 6 shots during the first 2 hrs after the start up of the IM, measured all the cavities and since there was no variability within shots and cavities, for the the remaining of production I was selecting randomly 5 cavities for my subgroups and then I used X and R charts.